Sign up to save tools and stay up to date with the latest in AI
bg
bg
1

Etched raises $120M in challenge to Nvidia in AI with transformer chips

Jun 25, 2024 - venturebeat.com
Etched, a company designing AI chips, has raised $120 million in a challenge to Nvidia. The company is developing a chip called Sohu that focuses on a critical part of AI processing: transform. By integrating the transformer architecture into its chips, Etched aims to create the world's most powerful servers for transformer inference. The funding round was led by Primary Venture Partners and Positive Sum Ventures, with participation from institutional investors including Hummingbird, Fundomo, Fontinalis, Lightscape, Earthshot, Two Sigma, and Skybox Data Centers.

Etched's Sohu chip is designed specifically for transformers, making it the fastest of its kind, according to the company. The firm believes that the future of AI will be dominated by transformers, and if this is the case, their chips could revolutionize the industry. However, if transformers are replaced by other architectures, Etched's chips will become obsolete. The company is partnering with TSMC to start production on their 4nm node and has already attracted engineers from major AI chip projects.

Key takeaways:

  • Etched, a company specializing in AI chips, has raised $120 million in a challenge to Nvidia. The company is designing a new chip called Sohu that focuses on a critical part of AI processing: transform.
  • The company believes that its team of 35 people can outperform Nvidia, and has received backing from investors such as Peter Thiel, Primary Venture Partners, and Positive Sum Ventures.
  • Etched's Sohu chip is designed specifically for transformers, making it the fastest transformer chip of all time. The company believes that if transformers continue to dominate AI models, their chip will revolutionize the industry.
  • Etched's founders, who dropped out of Harvard University, believe that the future of AI lies in scaling and specialization. They argue that as AI models become larger and more complex, specialized chips like Sohu will become inevitable.
View Full Article

Comments (0)

Be the first to comment!